How do I scale my waveforms with frequency?

I have real world waterfall FFTs with several very clear harmonic frequencies. I've written a function that takes an array with those frequencies and a scaling factor for each frequency. The function then sums scaled sinusoids at those frequencies to produce a 'wav' file. When I use scaling factors according to the intensity profiles on the waterfall diagrams, the high frequencies are FAR to loud, I get reasonably good simulations of the real world sounds if I just leave out the higher frequencies but I'd like to know how to scale them appropriately for a more accurate waveform. I've tried dividing the scaling factors by respective frequency, the square of the frequency, and several other things. I'm obviously not an acoustic engineer, but I've been trying to read up on sound power/pressure/intensity to see what I've got wrong, but I haven't been able to glean anything from the info.

tl;dr: When I look at a real-world waterfall FFT, and see two bands at comparable intensities, one at 150Hz, one at 1200Hz, what does that mean for the amplitude of their respective sinusoidal waveforms?